Search for: All records

The deployment of deep learning-based malware detection systems has transformed cybersecurity, offering sophisticated pattern recognition capabilities that surpass traditional signature-based approaches. However, these systems introduce new vulnerabilities requiring systematic investigation. This chapter examines adversarial attacks against graph neural network-based malware detection systems, focusing on semantics-preserving methodologies that evade detection while maintaining program functionality. We introduce a reinforcement learning (RL) framework that formulates the attack as a sequential decision making problem, optimizing the insertion of no-operation (NOP) instructions to manipulate graph structure without altering program behavior. Comparative analysis includes three baseline methods: random insertion, hill-climbing, and gradient-approximation attacks. Our experimental evaluation on real world malware datasets reveals significant differences in effectiveness, with the reinforcement learning approach achieving perfect evasion rates against both Graph Convolutional Network and Deep Graph Convolutional Neural Network architectures while requiring minimal program modifications. Our findings reveal three critical research gaps: transitioning from abstract Control Flow Graph representations to executable binary manipulation, developing universal vulnerability discovery across different architectures, and systematically translating adversarial insights into defensive enhancements. This work contributes to understanding adversarial vulnerabilities in graph-based security systems while establishing frameworks for evaluating machine learning-based malware detection robustness. 

Abstract El Niño/Southern Oscillation variability has conspicuous impacts on ecosystems and severe weather. Here, we probe the effects of anthropogenic aerosols and greenhouse gases on El Niño/Southern Oscillation variability during the historical period using a broad set of climate models. Increased aerosols significantly amplify El Niño/Southern Oscillation variability primarily through weakening the mean advection feedback and strengthening the zonal advection and thermocline feedbacks, as linked to a weaker annual cycle of sea surface temperature in the eastern equatorial Pacific. They prevent extreme El Niño events, reduce interannual sea surface temperature skewness in the tropical Pacific, influence the likelihood of El Niño/Southern Oscillation events in April and June and allow for more El Niño transitions to Central Pacific events. While rising greenhouse gases significantly reduce El Niño/Southern Oscillation variability via a stronger sea surface temperature annual cycle and attenuated thermocline feedback. They promote extreme El Niño events, increase SST skewness, and enlarge the likelihood of El Niño/Southern Oscillation peaking in November while inhibiting Central Pacific El Niño/Southern Oscillation events. 

Abstract The ability to control the electrode interfaces in an electrochemical energy storage system is essential for achieving the desired electrochemical performance. However, achieving this ability requires an in-depth understanding of the detailed interfacial nanostructures of the electrode under electrochemical operating conditions. In-situ transmission electron microscopy (TEM) is one of the most powerful techniques for revealing electrochemical energy storage mechanisms with high spatiotemporal resolution and high sensitivity in complex electrochemical environments. These attributes play a unique role in understanding how ion transport inside electrode nanomaterials and across interfaces under the dynamic conditions within working batteries. This review aims to gain an in-depth insight into the latest developments of in-situ TEM imaging techniques for probing the interfacial nanostructures of electrochemical energy storage systems, including atomic-scale structural imaging, strain field imaging, electron holography, and integrated differential phase contrast imaging. Significant examples will be described to highlight the fundamental understanding of atomic-scale and nanoscale mechanisms from employing state-of-the-art imaging techniques to visualize structural evolution, ionic valence state changes, and strain mapping, ion transport dynamics. The review concludes by providing a perspective discussion of future directions of the development and application of in-situ TEM techniques in the field of electrochemical energy storage systems. 

Abstract Increasing the number of organ donations after circulatory death (DCD) has been identified as one of the most important ways of addressing the ongoing organ shortage. While recent technological advances in organ transplantation have increased their success rate, a substantial challenge in increasing the number of DCD donations resides in the uncertainty regarding the timing of cardiac death after terminal extubation, impacting the risk of prolonged ischemic organ injury, and negatively affecting post-transplant outcomes. In this study, we trained and externally validated an ODE-RNN model, which combines recurrent neural network with neural ordinary equations and excels in processing irregularly-sampled time series data. The model is designed to predict time-to-death following terminal extubation in the intensive care unit (ICU) using the history of clinical observations. Our model was trained on a cohort of 3,238 patients from Yale New Haven Hospital, and validated on an external cohort of 1,908 patients from six hospitals across Connecticut. The model achieved accuracies of$$95.3~\pm ~1.0\%$$and$$95.4~\pm ~0.7\%$$for predicting whether death would occur in the first 30 and 60 minutes, respectively, with a calibration error of$$0.024~\pm ~0.009$$. Heart rate, respiratory rate, mean arterial blood pressure (MAP), oxygen saturation (SpO2), and Glasgow Coma Scale (GCS) scores were identified as the most important predictors. Surpassing existing clinical scores, our model sets the stage for reduced organ acquisition costs and improved post-transplant outcomes.